Rice, Minnesota scientists use predictive modeling to identify optimized zeolites to aid ethanol, petroleum production
Scientists at Rice University and the University of Minnesota have identified, through a large-scale, multi-step computational screening process, promising zeolite structures for two energy-related applications: the purification of ethanol from fermentation broths and the hydroisomerization of alkanes with 18–30 carbon atoms encountered in petroleum refining.
The results, presented in a paper published in Nature Communications, demonstrate that predictive modeling of synthetic zeolites—a technique pioneered by Rice bioengineer Michael Deem—and data-driven science can be applied to solve some of the most challenging problems facing industries that require efficient ways to separate or catalyze materials.
The resulting zeolites attack two complex issues. Current ethanol production is a multistep process involving fermentation of biomass. One of the last steps involves the separation of ethanol from water, and the researchers found an all-silica zeolite with pores and channels that accommodate ethanol molecules but shun hydrogen bonding with water molecules.
The second zeolite would improve the process of separating linear long-chain and short, branched hydrocarbons called alkanes, which affect the pour point and viscosity of lubricants and other petroleum products. Defining appropriate sorbents and catalysts for all of the complex mixtures involved in creating these products has been “an intractable problem,” the researchers wrote.
Deem is co-author of the new study led by University of Minnesota chemist Ilja Siepmann. Siepmann is director of the Minnesota-based Nanoporous Materials Genome Center, of which Deem is a member.
Zeolites are molecular sieves, nanoscale particles with channels and pores; they play numerous important roles in modern petroleum refineries and have the potential to advance the production of fuels and chemical feedstocks from renewable resources. The performance of a zeolite as separation medium and catalyst depends on its framework structure, the authors note in their paper. To date, 213 framework types have been synthesized and more than 330,000 thermodynamically accessible zeolite structures have been predicted.
The immense number of possible zeolites makes trial-and-error development nearly impossible. Accordingly, the successful identification of purpose-built zeolites for two applications is a valuable proof-of-concept for the algorithm, which Deem said has the power to accelerate the pace of materials discovery.
Deem suggested petroleum as a target for the Minnesota team. Refining and reforming petroleum is largely done with zeolites; the study catalytic hydroisomerization by zeolites to produce slightly branched, high-viscosity products, Deem said. The petroleum industry would see significant benefits from materials that improve chemical processes in the refining of crude oil, extending supplies and reducing costs, he added.
The candidate zeolites have yet to be synthesized, but the ability to screen millions of possibilities down to just a few with the right structures, pore sizes and chemical reactivity is a major accomplishment.
Now that the database has been screened for zeolites with these catalytic reforming properties, the materials we found become very interesting targets for synthesis. We’re looking for materials that have interesting properties, and that’s what we have achieved here.—Michael Deem
The Rice algorithm also suggests paths to synthesis of predicted zeolites.
Calculating the structures required serious computing power. The team used up to 800,000 processors on Mira, a supercomputer at Argonne National Laboratory, to tackle the problem.
Deem noted that labs create about five synthetic zeolites a year without resorting to the algorithm, but inevitably they are among zeolites predicted by it.
Basically, that says the heretofore unknown structures in our database, at least some of them, are synthesizable. We predicted them and they can be made.—Michael Deem
Co-authors of the new study are Pent Bai, Mi Young Jeon, Limin Ren and Michael Tsapatsis of the University of Minnesota, and Chris Knight of Argonne National Laboratory. Siepmann is the Distinguished McKnight University Professor and Distinguished Teaching Professor and the Merck Professor of Chemistry at the University of Minnesota. Deem is chair of Rice’s Department of Bioengineering, the John W. Cox Professor of Biochemical and Genetic Engineering and a professor of physics and astronomy.
The Department of Energy Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences and Biosciences, supported the research.
Peng Bai, Mi Young Jeon, Limin Ren, Chris Knight, Michael W. Deem, Michael Tsapatsis & J. Ilja Siepmann “Discovery of optimal zeolites for challenging separations and chemical transformations using predictive materials modeling” Nature Communications 6, Article number: 5912 doi: 10.1038/ncomms6912